Configure Retrieval Augmented Generation (RAG) with Amazon Bedrock Knowledge Bases and MongoDB

Amazon Bedrock offers a feature called Knowledge Bases, which allows you to connect Large Language Models (LLM) to additional data documents, such as PDFs, Markdown, HTML, Microsoft Word & Excel files, and more. This technique is also known across the industry as Retrieval Augmented Generation (RAG). Once you connect a model to your document store, in Amazon S3, you can query / prompt the model to answer questions or otherwise consume data that are stored inside of those documents. In order for this feature to work, you must connect Amazon Bedrock to a supported vector database. At the moment, Bedrock supports five different vector storage engines, including Amazon OpenSearch, Aurora Postgres, Pinecone, Redis Enterprise Cloud, and MongoDB. In this video, Trevor Sullivan (Solutions Architect, StratusGrid) explores setting up MongoDB Atlas as a vector storage engine, and connecting Amazon Bedrock Knowledge Bases to it.
Learn more at stratusgrid.com/

[@-SANJAIMI]

hello sir!!! I have a issue in creating the knowledge base. When I create it shows failed to create open search serverless collection.eventhough I gave the full access access for bedrock and opensearch service for the user and made the s3 bucket to be accessed by the opensearch service, the issues is not fixed. can you help me to clear that issue? I'm struggling with that issue !!! please help me !!!

[@k4rthikr]

is it possible to get links to the dataset that was used for this demo?

[@codeinrust]

We don't have copyright permission to distribute those materials. You can search for things on Google like "subaru legacy 2019 pdf manual," or other vehicles, to locate similar files for your usage.

[@MohitThehuman]

Great video
Can we connect a database like mysql rds etc… with bedrock ?

[@codeinrust]

Hello, thanks for your feedback. The only supported relational database engine is Amazon Aurora Postgres. Unfortunately you can't use MySQL or Postgres by themselves with Bedrock Knowledge Bases at the moment. You can provide feedback to the Bedrock team and ask them to support other vector databases though.

[@lucasbergamo7607]

for me fail the connection..